New framework could help diagnose recessive rare diseases

Many genetic diseases are ‘recessive’, meaning a person must inherit two faulty copies of a gene — one from each parent — to become ill. Carriers with just one faulty copy are usually healthy, which makes these variants difficult to identify and interpret.

In ADSL deficiency, a rare condition causing seizures, developmental delay and intellectual disability in children, nearly all patients carry two different variants, one in each copy of the gene.

Predicting whether a particular combination will cause disease is a major challenge, and computational tools that work well for other genes perform poorly for ADSL.

To address this, researchers at the Institute of Genetics and Cancer developed a large-scale experimental approach, testing over 8,000 variants of the human ADSL gene in yeast cells, which can use the human enzyme in place of their own. 

The key innovation was running the experiment twice: once with the gene expressed at a low level and once at a high level. Because cells respond differently to changes in enzyme activity depending on how much enzyme is present, comparing results across the two conditions allowed the team to mathematically extract estimates of true enzyme activity for each variant — information that is normally hidden within raw experimental data.

Armed with these insights, the researchers then tackled the real clinical question: what happens when two variants occur together in the same patient?

They developed a biallelic pathogenicity score, a method that combines the estimated activities of both alleles. This score not only matched the enzyme activity measured directly in patient samples but also proved more effective in distinguishing between affected and healthy individuals than existing computational prediction tools.